The SLAM Benchmark Framework is designed for evaluating 2D SLAM maps by comparing them with ground truth maps. This framework consists of pre-processing and evaluation stages, allowing users to assess the accuracy of various 2D SLAM algorithms.
-
Pre-processing Stage:
- Image Registration
- Thinning Operation
-
Evaluation Stage:
- Image Similarity
- Geometric Distance Measurement
- Correspondence Matching
- Python 3.x
-
Clone the repository:
git clone https://github.com/your-username/slam-benchmark-framework.git
-
Install dependencies:
pip install -r requirements.txt
-
Usage
Run the main UI script:
python main_ui.py
Use the UI to select the ground truth map and SLAM map.
Follow the steps in the Pre-processing and Evaluation stages as needed.
Click "Run Benchmark" to initiate the benchmarking process.
-
File Structure
main_ui.py: The main script for the graphical user interface. benchmark_logic.py: Contains the logic for image registration, thinning operation, and benchmarking. utils.py: Utility functions used in the project. ...
-
Contributing
Contributions are welcome! If you find any issues or have suggestions, please open an issue or create a pull request.
- License
This project is licensed under the MIT License.
- Contact
For inquiries, please contact riadh.dhaoui@rub.de